The Pilot Stroke Data Bank project has developed operational diagnostic algorithms for the classification of stroke type. These algorithms are based on the type of laboratory evidence available (CT, angiography, etc.), its findings, and the severity of stroke as measured by the neurologic exam. In this study, the usefulness of these diagnostic algorithms in differentiating etiology and predicting outcome is evaluated. The correlation of various factors relating to apparent etiology, stroke severity, and long-term prognosis is measured also.